Random Data Generator

Why should I use a fixed seed in my tests?

Test suites that generate random data without seeding produce different output on every run, which means a test that asserts on specific values either flakes randomly or has to weaken its assertions to 'any non-empty string' and lose its actual signal. Fixing the seed (passing a specific integer to the generator) guarantees the same random draw sequence on every run, on every machine, in CI. When a test fails, the reporter shows the exact row that failed, and you can reproduce locally with the same seed. This is the standard pattern in Jest, Mocha, Go's testing/quick, and Hypothesis — always seed, never hope.

Can I use the generated emails to sign up for free trials?

No. Every serious signup flow validates emails beyond surface syntax: MX record lookup, disposable-domain blocklists (ten thousand temporary-mail domains are well-known to every major anti-abuse service), Gmail-specific +tag heuristics, and increasingly pattern detection that flags suspiciously random-looking local parts. A generated email like jessica.martin.98342@example.com will fail MX validation immediately because example.com intentionally does not accept mail. For real signup testing, use aliases on a domain you own or your own catchall subdomain so the emails are deliverable to an inbox you can actually check.

Are the generated addresses real places?

City, state, and postal-code are drawn from real directories so they validate correctly in address-aware forms (USPS API, Google Address Autocomplete, European postal services). Street number and street name are synthetic — 1247 Maple Street exists in thousands of US cities in the abstract, but the specific combination 1247 Maple Street, Springfield, OR 97477 will almost never be an actual delivery point. That is the intended design: plausible enough to pass form validation, not real enough to risk accidentally generating a living person's home address and leaking it into a demo or a bug report.

How do I generate nested / structured objects?

Use the schema mode: define the shape as a JSON template where field values are replaced with generator tokens like {{firstName}}, {{email}}, {{address.city}}, {{date.past}}. The generator walks the template and replaces each token with a random value for each of the N records you request. Nested objects work the same way — a user object can contain an address sub-object with city, state, postal code as its own sub-fields, and each request produces a full tree with coherent locale-consistent data. Output as JSONL for large datasets or as an array for direct paste into a test fixture file.

What is the largest dataset I can generate in the browser?

Generation is CPU-bound but not memory-bound for reasonable record counts — 10,000 records complete in under a second on a modern laptop, 100,000 in 5 to 15 seconds depending on field complexity. Beyond 100,000 records the browser's heap starts to pressure with the full array in memory before export, especially for JSON output where the pretty-printed string can be several hundred megabytes. For larger datasets stream to CSV (lighter format, no stringification overhead) or run Faker directly in Node with a streaming writer so output is written to disk as it is generated rather than accumulated in RAM.